Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Randal Beard
For my final project I used a program developed by Mark Drela and Harold Youngren from MIT
to calculate the aerodynamic coefficients of two small fixed wing aircraft, namely a Pelican and Zagi
model. Athena Vortex Lattice (AVL) is designed to be used on wing designs that are thin and are subject
to small angles of attack. The program breaks up a specified lifting surface into discretized sections both
spanwise and chordwise. The program then applies horseshoe vortex calculations across these sections.
AVL also has the capability of modeling slender fuselages, though they are not required to calculate the
plane’s various parameters.
To begin my project, I first measured both plane’s physical dimensions which are listed below in
table 1. A note on the values: AVL doesn’t care about the units, just as long as you are consistent with
the magnitudes.
1
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
In calculating the parameters for the Zagi model, I found the web app shown in Figure 3 to be
extremely convenient. A link to the page is given in the Appendix.
AVL requires only one file to run, however, three types of input files are accepted for performing
tests. The first, and only required file, is the physical geometry as described in an x, y, z coordinate
frame. Instead of a north, east, down configuration, the axis reflects a downstream, right wing, up
position. The second file describes the mass and inertia distribution while the third input file is a run
case where you can initialize different test conditions rather than inserting them into the running
program. The proper text file format is given in the documentation available online through the link
given in the Appendix. For my tests, I only ran AVL with the geometry file which allowed me to calculate
the aerodynamic coefficients. With the mass distribution, you can run an Eigenmode analysis to gain
further information on the aircrafts design.
When inputting the geometric values, you need to supply a data point set describing the camber
of the airfoil. The University of Iowa has made available an extensive online database with short
descriptions of the general use of each airfoil listed. I found by trial and error that AVL requires that
these points be in a trailing-to-leading-to-trailing orientation to be drawn correctly in the geometry
2
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
viewer. You can check to make sure the orientation is correct by activating the camber lines (“CA”
command) while in the geometry viewer. Figure 4 shows a profile of the camber lines.
Once the geometry is loaded into AVL, you can begin tests by typing “OPER”. Type “x” to run an
initial test with test conditions all set to zero. Test conditions include elements such as control surface
angles of deflection, angle of attack or sideslip, and various other parameters. Once the calculations are
done, typing “g”, then “lo” will show the loading on the lift surfaces. An example is shown in Figures 5
and 8. The leading edge can be seen to provide a negative load, whereas the rest of wing area provides a
positive load producing lift. In Figure 6, the angle of attack was changed to 3 degrees and the
calculations were run again by typing “x”. The leading edge has significantly reduced negative load while
the winglets provide a lateral load toward the center of the plane, adding stability. It can also be seen
that a positive load is applied to elevons.
Running the calculations again with the angle of attack set back to zero and then typing “st” will
show the stability derivatives. These can be printed to a file, or directly onto the terminal. Both the
Zagi’s and the Pelican’s data are presented below. Figures 7 through 9 show the same run settings as the
Zagi, but for the Pelican.
This AVL program has some learning curve to it, but proves to be extremely useful once the
format is figured out. I was able to find a paper done by a group in Australia that compared AVL results
to actual wind tunnel testing of their small airframe. In the paper’s results, they showed that AVL is
more accurate than a full 3-d panel derived calculation program called PanAir. I feel that AVL will be a
good addition to the research going on not only in the 674 class, but in the MAGICC lab as well. I was
only able to explore a few of the capabilities of this program, but I hope future students of the class can
learn more from it.
3
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
4
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
5
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
Summary of results
An important note: According to the tutorial, AVL calculates each stability derivative in terms of per
radian, however the control surface derivatives are in per degree. The values given in the tables below
are converted to radians as is required in the equations calculated in the class book (top of page 45).
Additional confirmation of the values for the control derivatives should be made. The values for 𝐶𝐷𝛼
and 𝐶𝐷𝑞 were not directly available through AVL that I could tell and I could not find a reliable
method for calculating them. Also, the values for the pelican’s 𝐶𝑌𝛿𝑟 and 𝐶𝑙𝛿𝑟 seem to me to be
incorrect, however, I could not find the source of the problem.
6
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
7
Andrew Hendricks ECEn 674 Final Project description Dr. Randal Beard
Appendix
MIT AVL main page
http://web.mit.edu/drela/Public/web/avl/
http://web.mit.edu/drela/Public/web/avl/avl_doc.txt
http://m-selig.ae.illinois.edu/ads/coord_database.html#E
http://www.flightlab.net/Flightlab.net/Download_Course_Notes_files/1_%20AxesDerivatives.pdf
ICAS paper
http://www.icas.org/ICAS_ARCHIVE/ICAS2012/PAPERS/675.PDF